|
Fri Jan 4 16:08:12 EST 2008
did HAL malfunction?
--
I finally got around to watching, 2001: A Space Odyssey,
and while most of the movie's point escaped me, I did come out of
it with at least one conclusion: I don't think the HAL 9000 com-
puter malfunctioned, which makes their claims of the computer be-
ing "error free" true even though the computer seemed to have
gone crazy.
First, I need to point out that I believe it is impossible for a
computer like HAL to be created by humans. That being said, I am
going to suspend that belief so I can argue that HAL did not mal-
function - given the technical achievement in the movie was a re-
ality.
When the trip to Jupiter sequence was introduced, it was hammered
home that the computer was build to simulate a human brain. It
was also mentioned that some scientists believed that the comput-
er merely mimicked the human brain, and I think that point was
raised to bring up the question I am addressing right now.
Wikipedia defines psycopathy as, classified by some as a per-
sonality disorder, characterized by amoral and antisocial behav-
ior. It is a term derived from the Greek psyche (soul, breath
hence mind) and pathos (to suffer), and was once used to denote
any form of mental illness, often being confused with psy-
chosis.
While HAL was clearly acting in a psycopathic manner, it was
still behaving as a psycopathic human. It can be said that a psy-
copathic human might be malfunctioning, but since no bounds were
set on which human HAL was simulating, we can not say that
it was failing to meet its design requirements.
Another interesting aspect to HAL is that it claims to have been
"turned on" in Urbana, Illinois during the year 1993. Coinciden-
tally (or not very) Urbana is home to UUIC, which has a very
strong computer science department. It is home to NCSA, and is
at the forefront of high performance computing research. It is
no stretch of the imagination to me that something like HAL would
be build there (i.e., if I though this was even possible). I di-
gress.
When HAL was "born," it recalls that its creator taught him a
song (called, "Daisy", I believe). This hints to the idea that
HAL's intelligence was grown organically, and it was taught and
raised as a human child might be. This is not unlike the theme
of many other AI movies - or research. This implies that, com-
pared to the complexity of its learned higher order cognative
functions, the underlying technology and circuitry required to
facilitate learning/searching/associating would not be overly
complex.
This also implies that its intelligence was not explicitly de-
signed, and it therefore became virtually impossible to know what
it would do, how it would react, etc. In otherwords, it would
have been just as impossible to read its mind and predict what it
would do as it is to do the same with a human mind (or neural
net). Furthermore, the only way to determine what its reaction
would be to a given stimulus would be to observe it react in real
time. This means prediction would be impossible, and while sta-
tistical models of its behavior might have been able to be con-
structioned, it would be no more possible to predict its reaction
based on a series of similar stimuli than it would to predict how
a real person would behave to a series of similar stimuli. Cou-
ple this with the fact that HAL was clearly a psycopath, and you
get a system that you can not predict. This does beg the ques-
tion, if you can't predict how a system would be have why claim
it is error free? However, it must be made clear - what system
are you saying is error free? The electronics and basic system
software - or the orgically grown behavior model that drives much
of its actions?
Who knows what happened to HAL in order to make it so paranoid
and crazy. One thing is pretty obvious - that it was acting out
just like a human with a similar psycopathic condition would have
given the situation.
Another thing I noticed, which would indicate an irrational core
of being inside of HAL, was that instead of taking decisive ac-
tion to protect itself from being dismantled it simply tried to
convince Dave to not do so.
If any part of HAL's behavior could have been considered an er-
ror, it would have been its lack of self preservation. Then
again, who is to say that a psycopath even feels this instinct if
it doesn't feel other instincts or emotions?
There is no stong evidence to me that HAL did indeed malfunction.
To the contrary, I feel like the computing system worked as de-
signed. The emergence of its psycopathic condition can not be
readily attributed to electronics or basic system programming.
One can, however, easily show that HAL was accurately and con-
vincingly simulating the actions and behavior of a truly psyco-
pathic personality. Perhaps the only human induced error of the
system was not properly conditioning or raising HAL or its twin
(mentioned midway through the sequence).
--
Powered by vee
Copyright © 2006-2008
|
|